Here are the key steps for Exercise 3:
1. Create a FileDataModel object, passing in the CSV file
2. Instantiate different UserSimilarity objects like PearsonCorrelationSimilarity, EuclideanDistanceSimilarity
3. Calculate similarities between users by calling userSimilarity() on the similarity objects, passing the user IDs
4. Print out the similarities to compare the different measures
The CSV file should contain enough user preference data (user IDs, item IDs, ratings) for the similarity calculations to be meaningful. This exercise demonstrates how to easily plug different similarity functions into Mahout's common interfaces.